Tag Archives: compositional hierarchical model

A different white-box deep architecture? Is it possible?

I recently published a paper more thoroughly explaining what I have been developing as a part of my PhD research at the Laboratory for Computer Graphics and Multimedia, Faculty of Computer and Information Science.

The paper describes an alternative deep architecture. Those of you who are involved in recent machine learning research probably know neural-network-based deep architectures, which recently gained a lot of attention from general public due to their results in a variety of fields. However, they are not perfect (at least I think so … :) ). To put it in a simple description, these networks lack transparence. We know how they work, but cannot see what they learn and thus often use them as black boxes (provide input, get output, don’t care about the process). We propose a different kind of deep architecture based on compositionality (large complex things are built of small simple things).

We tested our model on music information retrieval tasks and are currently seeking problems in other scientific fields dealing with big data – please message me for suggestions or potential collaboration.

If you are interested, I welcome you to read our paper, (freely) available here: http://dx.doi.org/10.1371/journal.pone.0169411

For a quick overview, I attached the paper’s abstract below:

Robust Real-Time Music Transcription with a Compositional Hierarchical Model

The paper presents a new compositional hierarchical model for robust music transcription. Its main features are unsupervised learning of a hierarchical representation of input data, transparency, which enables insights into the learned representation, as well as robustness and speed which make it suitable for real-world and real-time use.

The model consists of multiple layers, each composed of a number of parts. The hierarchical nature of the model corresponds well to hierarchical structures in music. The parts in lower layers correspond to low-level concepts (e.g. tone partials), while the parts in higher layers combine lower-level representations into more complex concepts (tones, chords). The layers are learned in an unsupervised manner from music signals. Parts in each layer are compositions of parts from previous layers based on statistical co-occurrences as the driving force of the learning process.

In the paper, we present the model’s structure and compare it to other hierarchical approaches in the field of music information retrieval. We evaluate the model’s performance for the multiple fundamental frequency estimation. Finally, we elaborate on extensions of the model towards other music information retrieval tasks.

Full paper is available here: http://dx.doi.org/10.1371/journal.pone.0169411

Enjoying the tutorial on automatic transcription. In a few years, CHM will overtake all the current approaches :)

ISMIR 2015 – Malaga – Day 0 & 1

Another year, another ISMIR. This time, we have submitted 3 posters – two for the Mirex evaluation – Urša and I adjusted the Compositional Hierarchical Model (CHM) to find re-occurring patterns in symbolic data (MIDI), Matija submitted a classifier for Music/speech discrimination; and a Late Breaking Demo poster for WebCHM – a web-based version of the CHM which is becoming the ultimate tool for transcription and evaluation.

As always, I will post few pics and tricks about the conference.

First flight, over lake bled

First flight, over lake bled

Second flight, somewhere over Spain

Second flight, somewhere over Spain

After we landed, we went around the town to familiarize ourselves with all the local places.

Malaga version of Ljubljana

Malaga version of Ljubljana

Something smells fishy around here ... ;)

Something smells fishy around here … ;)

The main church and its default "spanish" look - one of a few places which makes you feel you are in Spain

The main church and its default “spanish” look – one of a few places which makes you feel you are in Spain

Vamos a la playa! playlist selection: White pearl black oceans

Vamos a la playa! playlist selection: White pearl black oceans

Wait! Where is everybody?

Wait! Where is everybody?

The only visitor who doesn't mind taking picture at a close range.

The only visitor who doesn’t mind taking picture at a close range.

Otherwise, some buildings in the town look pretty interesting. No thermal isolation,  no electrical standards.

Otherwise, some buildings in the town look pretty interesting. No thermal isolation, no electrical standards.

We stopped for a research meeting. The food here is interesting, the beer … not so much.

On the other hand, we found the troll :)

Best costume award

Best costume award

During the first day, we visited the technical department of the local university. It was raining and since the roads have no drain shafts, there was a significant amount of water on the road. Trick question: which registration plate is in the following picture?

The taxi driver has no intention to use the wipers. He can see just fine!

The taxi driver has no intention to use the wipers. He can see just fine!

The university building is HUGE, here is a picture of one of the stairway, easily accommodating  hundreds of students.

Left side of the stairway (it seems smaller due to the zoom of the camera)

Left side of the stairway (it seems smaller due to the zoom of the camera)

I’m finishing this entry with a picture of the current tutorial.

Enjoying the tutorial on automatic transcription. In a few years, CHM will overtake all the current approaches :)

Enjoying the tutorial on automatic transcription. In a few years, CHM will overtake all the current approaches :)

MusicTechFest – Compositional hierarchical model

Today, Manca, Luka and I presented the Compositional hierarchical model at the Music of Science session at Music Tech Fest in Ljubljana. As a part of his undergraduate degree, Luka built an awesome transcription editor which uses the CHM as classifier. It will probably never reach version 1.0 since the “ideas” stack grows exponentially. However, we are planning on providing a freeware version shortly.

Manca showed how the CHM can be used for music generation – aka. how to build a hammond we can’t afford :)

The CHM - my PhD in under 2 minutes

The CHM – my PhD in under 2 minutes

Let's play the CHM like a hammond!

Let’s play the CHM like a hammond!

The CHM transcription editor

The CHM transcription editor